# Thread: Sample size for a set of correlations averaged into pearsons r

1. ## Sample size for a set of correlations averaged into pearsons r

Hi all

I am conducting a meta analysis of pearson r correlations. I have converted these into fishers z to analyse in SPSS. Some studies presented more than one correlation , but measuring the same construct with the same method. I have averaged these correlations via fishers z but I now have a problem in that I need to calculate a standard error and the deviance which , I believe , requires one sample size per correlation.
I am wondering if it is acceptable to calculate an average of the sample size by combining the samples and dividing by the number of samples please?

2. ## Re: Sample size for a set of correlations averaged into pearsons r

Hey laboh.

Hint - Consider using central limit theorem and the observables you use for data are correlation estimates - not normal observables. [i.e. xi = correlation_i].

This is if you have lots of estimates and we assume IID.

3. ## Re: Sample size for a set of correlations averaged into pearsons r

Hi Chiro

Thanks. There are 29 correlation estimates in total ( when all are averaged - i.e 29 studies with one correlation). There are two studies with two separate correlations and sample sizes , there is one study with three separate correlations and sample sizes.